Theft of an Idol: Text and Context in the Representation of Collective Violence

By Paul R. Brass

As collective violence erupts in many regions throughout the world, we often hear media reports that link the outbreaks to age-old ethnic or religious hostilities, thereby freeing the state, its agents, and its political elites from responsibility. Paul Brass encourages us to look more closely at the issues of violence, ethnicity, and the state by focusing on specific instances of violence in their local contexts and questioning the prevailing interpretations of them. Through five case studies of both rural and urban public violence, including police-public confrontations and Hindu-Muslim riots, Brass shows how, out of many possible interpretations applicable to these incidents, government and the media select those that support existing relations of power in state and society.


Adopting different modes--narrator, detective, and social scientist--Brass treats incidents of collective violence arising initially out of common occurrences such as a drunken brawl, the rape of a girl, and the theft of an idol, and demonstrates how some incidents remain localized while others are fit into broader frameworks of meaning, thereby becoming useful for upholders of dominant ideologies. Incessant talk about violence and its implications in these circumstances contributes to its persistence rather than its reduction. Such treatment serves in fact to mask the causes of violence, displace the victims from the center of attention, and divert society's gaze from those responsible for its endemic character. Brass explains how this process ultimately implicates everyone in the perpetuation of systems of violence.

• Language: English
• Publisher: Princeton University Press
• Release date: Sep 1, 2020
• ISBN: 9780691217918

Book preview

Chapter 1

TEXT AND CONTEXT

IN I, Pierre Rivière, Foucault presents the text—the memoir of a parricide and the dossier associated with it—as a case, an affair, an event that provided the intersection of discourses that differed in origin, form, organization, and function. The discourses of judges, prosecutor, country doctor, the great psychiatrist Esquirol, the villagers, including their mayor and parish priest, and the murderer himself all appear[ed] to be speaking of one and the same thing, of the murders that occurred on June 3, 1835, but the multiplicity of discourses form[ed] neither a composite work nor an exemplary text, but rather a strange contest, a confrontation, a power relation, a battle among discourses and through discourses, used as weapons of attack and defense in the relations of power and knowledge.¹ This contest was one for the definition of the truth of the parricide’s act, the truth not of its commission, for which the murderer’s own confession provided sufficient evidence, but the truth of its explanation and meaning, for the context in which it should be placed, and for its broader implications concerning the workings of the human mind in an age of reason from which madness could not be banished.

This struggle for the appropriate explanation of the parricide’s act and for the choice of the appropriate punishment as well involved a contest for domination within domains of legal, medical, criminological, and psychiatric truth and within the institutions in which those domains of truth and knowledge were generated: courts, prisons, hospitals, asylums. It also brought clearly into view the gap in local and national systems of knowledge and explanation. The local doctor, unable to find scientific categories in which to place and define the parricide and his act, saw no reason for the intervention of medico-psychiatric knowledge into this case, which he thought should be left to the judicial system. The Parisian psychiatrists, of course, ultimately succeeded in establishing their claim and the consequent spread of psychiatric knowledge not only into the judicial and penal system but throughout modern Western society.

Another consequence of their ultimate victory was the removal from local societies of the power to define and interpret local incidents of violence, to place them in a specific context based on local knowledge. Knowledge of human behavior in nineteenth-century France and in the rest of Western society became nationalized and universalized. Events that occur in isolated villages and hamlets or on the city streets have become subject to placement in categories and contexts previously not known to or incidental to the lives of those who experience them. People with personal knowledge at the sites of occurrences of violence, lacking knowledge of the appropriate scientific categories in which to place them or refusing to accept the contextualizations of them imposed by outsiders, continue to generate their own interpretations. The more perfectly integrated contemporary societies become, however, the more local and national interpretations merge and the more persons in the localities find themselves also readily placing local events into ready-made categories and contexts.

In the twentieth century, local acts of violence are often also commonly placed in other, more direct kinds of political contexts depending upon the ethnic identities of the persons involved in them. Interethnic relations have become such a pervasive concern in contemporary societies that the interpretation of virtually any act of violence between persons identified as belonging to different ethnic groups itself becomes a political act. But the interpretive process is not only political; it also generates competing systems of knowledge concerning interethnic relations, the sources of tension between members of ethnic groups, the causes of discrimination and prejudice, their social and economic bases, and the like. The developers of these systems of knowledge generally claim to stand apart from and above the ethnic interactions they seek to explain and to describe objectively, whereas in fact they themselves play a role in their perpetuation by the very process of placing interethnic incidents of violence in broader political contexts.

Interethnic incidents of violence in modern states invariably involve other parties in addition to those who contextualize them in journalistic or social scientific accounts, notably the authorities and especially the police. The police often prefer to define local incidents of violence, whether between members of different ethnic groups or not, simply as crimes and to treat them as such, that is, to localize and confine them. Such localization, however, becomes more difficult when isolated incidents of interethnic violence become transformed into something broader, a riot involving large numbers of massed persons from opposing ethnic groups engaged in assaults on persons, lives, and property. In such situations, the police may themselves take sides; even when they do not, they are often accused of doing so, particularly by the side that emerges with the greater losses. Some ethnic riots do not even involve two ethnic groups, but often only one community and the police, with the latter perceived or described by spokesmen for the rioting community as responsible for the precipitating incidents leading to the riot. When it can be proved that the police and the state authorities more broadly are directly implicated in a riot in which one community provides the principal or sole victims, then, of course, one is confronted with a pogrom.

The definition of occurrences of violence involving members of one or more ethnic groups as being merely isolated incidents having no broader significance or as being riots or pogroms is frequently itself a part of the political struggle. The boundaries between isolated incidents and proven pogroms are clear enough. However, those between local incidents of violence and riots, on the one hand, and between riots and pogroms, on the other hand, are not. Some so-called riots are no different in origin from everyday incidents of violence. Others, however, may be pogroms in disguise.

Interethnic or police–people confrontations involving violence in contemporary states almost always lead to the construction of interpretations of them by authorities, media, politicians, and political activists. The struggles to gain acceptance for particular constructions of violent and riotous behavior are inherently political, with important consequences for state policies and resource distribution.² The constructions that become officially or broadly accepted are usually far removed from the actual precipitating incidents and from local interpretations of them. The scholar who inserts himself into the description of such incidents in a search for objective truth can do no better than to place himself on one side or another in the political struggle. When the search for categories, constructs, and contexts is central to the political struggle, the scholar’s own concepts and the scholar himself cannot stand apart from and outside of them.

Where, then, can a scholar interested in questions of violence place himself in relation to these constructions? One possible posture is to seek to expose the falsifications contained in all of them. However, if one starts with the premise that violence in which innocent persons are harmed and killed is an evil, such a rhetorical strategy provides a poor vantage point. One must, therefore, take a stand rather in relation to the whole process of construction and contextualization. It is my purpose in this volume to find a place to stand in relation to the contextualization of incidents of violence in contemporary Indian society and to ask whether some kinds of constructions contribute more and others less to their perpetuation.

THEORY AND METHOD

The focus of this book is on issues of ethnicity and violence and their relationship to state-building and national unity in India. It is based on field research carried out between 1982–83 and 1994 and on incidents that occurred between 1980 and 1994 at sites of police–public and interethnic violence in north India, which in recent years has given all appearances of undergoing the most extreme paroxysms of caste, communal, and other forms of violence that appear to threaten the unity of the country and the stability of the Indian state. The original research for this project in 1982–83 was carried out with no other intent at the time than to engage in the social scientist’s conventional search for the truth of contemporary public violence through the use of comparative case studies, one of the accepted methods of the social science disciplines. A change in my design came afterward as I looked over my interviews in new ways, treating them as texts subject to multiple interpretations rather than as sources of valid or confirmed information or data about events.

I raise three questions concerning the increased incidence and intensity of violence in India in recent years. The first concerns the relationship between issues of ethnicity, communalism, and violence, on the one hand, and national unity, on the other hand. To what extent does this upsurge in violence and the specific forms it has taken indicate a serious threat to the continued unity and integrity of the country and the relations among its major peoples? Since questions of intercaste relations and even more of Hindu–Muslim relations are at the heart of much of this upsurge in violence, the second question is why caste and communal riots persist in India. Third, why especially have Hindu–Muslim riots become endemic? These three questions, I will show, are interrelated.

The argument I develop is that the publicized versions of many so-called caste and communal riots in India, like many aspects of ethnic identity itself, are constructions upon events that are usually open to a multiplicity of interpretations. When examined at the actual originating sites of ethnic and communal violence, it is often the case that the precipitating incidents arise out of situations that are either not inherently ethnic/communal in nature or are ambiguous in character, that their transformation into caste or communal incidents depends upon the attitudes toward them taken by local politicians and local representatives of state authority, and that their ultimate elevation into grand communal confrontations depends upon their further reinterpretation by the press and extralocal politicians and authorities. The official interpretation that finally becomes universally accepted is often, if not usually, very far removed, often unrecognizable, from the original precipitating events. This is particularly the case with those riots that are labeled as communal or Hindu-Muslim riots. Indeed, I intend to demonstrate that the persistence of riots so labeled is in large part a consequence of their functional utility for all dominant political ideologies, both secular and communal, in contemporary India. Their persistence helps local, state, and national leaders of different ideological persuasions in capturing or maintaining institutional and state power by providing convenient scapegoats, the alleged perpetrators of the events, and by providing as well dangers and tensions useful in justifying the exercise of state authority.

The approach adopted here is quite different from the conventional social science accounts of riots and their explanations, which are broadly of two types. The first seeks to develop from descriptions of riots at different times and places an abstract, objective model of their sequential character. The classic sequence in the literature on collective violence describes a progression from the juxtaposition of opposed ethnic groups or one ethnic group and the police to the creation of an atmosphere of tension arising out of grievances, conflicts, and frustrations followed by a triggering incident, which then produces a conflagration as does a spark upon some combustible material.³ This sequence is common also in popular and official accounts of the development of riots in India.

The second general approach to the explanation of riots, particularly in societies in which they are endemic or occur in waves, discards description and dynamic analysis for ecological comparison, seeking to identify the demographic, economic, and/or political conditions that distinguish cities and towns in which riots have occurred from those in which they have not. Such analyses usually produce tables and charts and ultimately regression equations that account for a percentage of the variance ranging most often from 25 percent to 50 percent, which is considered quite good in social science statistics. Thus, for example, a conclusion may be reached that 40 percent of the riots in America or in India have occurred in cities and towns where blacks and whites or Hindus and Muslims, respectively, live side by side in a 40–60 ratio or some other proportion. Other characteristics are then added to the regression equation that may increase the variance explained to 50 percent or sometimes even more. However far one goes, a substantial amount of the variance always remains unexplained.⁴ When one keeps in mind that large numbers of people are being killed in those cities and towns where the variance is unexplained, such forms of analysis provide cold comfort even if there were some hope that the explanations of social scientists could be translated into corrective public policies.

I have three objections to these alternative approaches. The first is that the claim of objectivity is unsustainable. No academic accounts of riots I have read are free of subjectivity. Second, these accounts, particularly the statistical, but also the classical sequential model, eliminate agency and responsibility from their explanations. They objectify a sequence or a demographic situation and fail to identify specific persons, groups, organizations, and state agents, who actually inaugurate and sustain riotous events and commit the arson, property destruction, and murder. Third, they fail to identify clearly the linkages between individual and social responsibilities and those of the state and society. Insofar as they do so, they tend to refer vaguely to the existence in the minds and hearts of the people of prejudices and antipathies from which in turn state actors are not free. Such statements have a pious ring to them, but they constitute smokescreens that mask persistent forms of speech, social practice, and organizational activities and the ways in which they form a net of complicity that may spread widely in societies.

In the research on the case studies for this volume, I found it impossible to present an objective account that explained each case or all of them fully, that is, which accounted for the event as a whole. Instead, I found more or less persuasive and unpersuasive, more or less scrupulous or self-serving, more or less dispassionate or prejudiced accounts, some of which contained elements of a whole truth of the event, but none that could explain the whole. Nor could I free myself completely from my own Western rationalist skepticism, social science training, and personal biases in seeking to construct a holistic account. At the same time, it was also evident that the riotous events I have here documented were not primarily spontaneous occurrences or chance happenings and that there were identifiable culprits who had conspired to produce or who had committed acts either designed to produce or whose effects were to produce riotous and murderous results.

Social scientists are trained to be skeptical of conspiracy theories and to leave the attachment of blame to journalists and the judicial system, but this has too often led to the objectification of social processes and the reification of categories such as class, ethnicity, or even human nature in a fruitless search for ultimate causes that ignores the dynamics of events, the significance of the interpretations—more or less tendentious—which are presented to explain them, and the uses to which the events and their explanations are put. I have in my accounts of riotous events in northern India, therefore, chosen to focus on neither the remote nor the precipitating causes of them, but on the action that occurs in the intermediate space and time, the interpretations that follow after the events, and, in the case of Hindu–Muslim riots, the existence of a grand interpretive framework, which provides for many an automatic explanation for any riot that can be traced to a triggering incident involving one or more Hindus and Muslims.

In some areas of India, such as the western districts of Uttar Pradesh, it is accepted journalistic and scholarly wisdom that virtually any incident between, say, a Hindu and a Muslim can easily lead to a flare-up and a communal riot. Does this, therefore, not reflect the latent hostility between the opposing communities, which requires only a spark to ignite it? How can such group antagonisms be considered social constructions? While it would be foolish to deny the existence of such hostilities in these areas, it is equally foolish to talk about latent hostilities as if all members of the opposing groups are poised to commit acts of violence.

Although most people everywhere are capable of committing acts of violence under a variety of circumstances, the kinds of violence that are committed in communal riots are, I believe, undertaken mostly by specialists, who are ready to be called out on such occasions, who profit from it, and whose activities profit others who may or may not be actually paying for the violence carried out. Such regions have developed what I call institutionalized riot systems, in which known actors specialize in the conversion of incidents between members of different communities into communal riots. Even here, however, not every such incident is allowed to develop into a riot. When full-fledged riots develop, the local politicians and authorities are often either incompetent or they themselves desire the riots to take place, and are willing to place a communal interpretation on the precipitating incidents.

The phrase institutionalized riot systems requires some elaboration and needs to be related to contemporary social science theories concerning the origins and development of movements of collective action and violence. Riots occupy a marginal and ambiguous place in contemporary theories of collective action, despite the work of historians such as E. P. Thompson, Natalie Zemon Davis, George Rude, and Eric J. Hobsbawm,⁵ and the outpouring of social science analyses of the Black ghetto riots of the late 1960s in the United States.⁶ They fall out of the definition of social movements as defined by Charles Tilly and used by him, Sidney Tarrow, and others, namely, as sustained and contentious forms of collective action.⁷ Clearly, riots are contentious forms of collective action, but they are not sustained. Their very episodic character, their apparent unpredictability, and the relative brevity of the time period in which they commonly occur have made them far more difficult to study than social movements involving large numbers of people concerting together for long periods of time. By the time the social scientist arrives on the scene, the riot exists only in memory and interpretation and its principal actors, some of whose activities, if revealed, would entitle them to a death sentence or life imprisonment, are either not to be found or are born liars.

The great outpouring of literature on the Black riots of the 1960s has led to some integration of riot theory into the general body of theory on collective action, but at some cost in the understanding of riots in their particularity. Most of the early literature on these riots fell into the categories already discussed above, namely, the sequential and ecological. The most imaginative writing on the subject, however, that by Doug McAdam, made a major contribution to collective action and collective movement theory in general by relating the Black riots to the history of Black protest and interpreting the riots as extensions of a historic insurgent movement.⁸ McAdam’s contribution in this regard was to develop a political process model to replace the older sociological sequential and ecological approaches. However, this model does not help us interpret particular riots either in the United States in the 1960s or elsewhere because it eliminates the question altogether of why a riot occurred in one city and not another, why it was more severe here than there, what interests were served in particular incidents, and how they relate to the broader goals of Black insurgency.

A further problem in the literature on riots and its relationship to the theoretical literature on collective action is the shift in the historical meaning of riots. After all, the term riot has historically meant a disorderly action that constitutes a direct assault upon or represents a danger to established authority. Even where riots were directed against groups other than the state, they were often considered assaults upon the state’s order or upon groups entitled to state protection. Thompson and Davis have separately argued that food riots in eighteenth-century England and religious riots in sixteenth-century France in effect constituted encroachments on state authority by groups who saw themselves as acting in place of the state.⁹

In some cases, such as nineteenth-and early-twentieth-century Russia, a specialized term, pogroms, was used to describe riots in which Jews were the main victims. This term, however, has now entered the theoretical literature on the subject to mean state-instigated and -supported riots against a minority ethnic group. In the twentieth century, the term riots continues to refer to challenges to the authority of the state, but those that draw the most journalistic and scholarly attention are ethnic or interethnic in form. In either case, riots do conform in part to the Tilly–Tarrow definition of social movements as involving collective action against the state or other groups, though, as already mentioned, they fail on the grounds that they are not sustained.

From the point of view of the questions raised in this volume, however, it is less important to find a definitional place for riots in theories of collective action than to consider the question of the changing meaning of riots and the scope the historical change has offered to the state authorities for dealing with riots. Why does the modern state, particularly the developing state such as India, see some kinds of riots as assaults on its authority and others not? Why are interethnic riots usually not so considered? How is it that state leaders may even find riots useful in justifying their authority?

For, indeed, some riots in India, notably communal riots, as I have already suggested, have larger political uses. They are used by state and national Muslim politicians to mobilize the Muslim minority, on the one hand, and by militant Hindu nationalists to consolidate Hindu communal sentiment, on the other hand. They also have political uses for secular nationalists. They provide a justification for their argument concerning the need to create a composite nationalism and a strong, united state. Secular nationalists also use riots for their own purposes of mobilizing the minorities. Riots also provide a useful smokescreen to divert attention from the demographic and economic context in which riots take place—in filthy slums unfit for human habitation whose habilitation does not fit into their economic development plans and designs for the transformation of India into a great and powerful industrial-military state.

RIOTS, MOVEMENTS, AND THEORIES OF COLLECTIVE ACTION

Riots, however much they differ from social movements, are nevertheless forms of collective action. Moreover, as just noted, some theorists, notably McAdam, but others also who wrote of the Black ghetto riots of the 1960s in the United States, even went to the extent of integrating those riots into the broader, historical social movement of what they described as Black insurgency or else treated them in the manner of Davis, Hobsbawm, Rude, and Thompson as basically structured, purposeful, rational, and politically meaningful.¹⁰ This theoretical move was made possible largely because the riots of the 1960s, like some of the other movements that preceded them, such as the lunch counter sit-ins, occurred in waves. Such waves or chains of riots have also occurred in other places at other times—the pogroms against the Jews in fourteenth-century Spain¹¹ and nineteenth-century Russia¹² being notable examples, as well as the post–World War I and –World War II racial riots in the United States, and Hindu–Muslim riots in India in the 1920s at the time of partition and in the aftermath of the destruction of the mosque at Ayodhya in 1992, among many other examples.

Riots, therefore, may occur in isolation or may follow upon one another in rapid succession. Riot analysts differ, however, on the significance of such waves. Those who see riots as spontaneous expressions of mass anger and resentment against the state or another group tend to see such waves or chains as feeding upon the news or rumors from the initial riot or riots, leading to acts of retaliation by the victims in one city or town upon the members of the other ethnic group alleged to have been the attackers in the preceding riots. Those who see riots as planned and orchestrated events, in contrast, are likely to see such waves or chains as part of an overall plan being implemented by a political party, the state, or, in the view of the state authorities, outside agitators.

There is yet a third possibility (which is the principal subject of this volume), namely, the fitting of the isolated riot into a broader framework. In this case, even a single riot may be transformed into something bigger and more meaningful through interpretation after the event. However, my argument goes beyond this point to encompass in the same process waves of riots as well. That is, I do not accept the easy integration of those riots that occur in waves or chains as either instances of insurgency or coordinated party or state action. In some cases, the argument is not only plausible, but obvious, as in Nazi Germany during Krystallnacht. In other cases, however, waves of riots may provide opportunities for a multiplicity of interests and purposes to be satisfied and expressed under the cover of the claim or interpretation of insurgency or communalism. Waves of riots may also occur as by-products of mass mobilizations in which incidents of violence break out in a number of places in the aftermath of provocative actions taken by participants in a broad social movement.

Such incidents, however, may or may not actually have something to do with the mass mobilization in question, that is, they may be part of the design of the movement or at least partly anticipated consequences of its leaders or they may take place under the cover of a movement to vent personal grievances or gain economic advantage. In the former case, it is important to consider the strategy of riots just as one considers the strategy of those who plan collective movements such as strikes, marches on a nation’s capital, or other forms of mass demonstration or civil disobedience. Even where riots are merely anticipated consequences of a mass movement rather than part of an overall design, the effects may be analyzed as if they constituted a strategy. The strategy or effects of a wave of riots and even of some that occur in isolation include intimidation of another group; protection of one’s own; the promotion of a myth concerning the violent intentions of another group by provoking its members into violent actions; the interpretation of the events afterwards to cast blame upon and discredit rivals, another group, or another group’s or party’s support base.

Further difficulties in incorporating riots into general theories of collective action arise when one considers that riots also stand in an ambiguous relationship to both routine and movement politics. In most civil societies, riots are considered both by the authorities and by elite public opinion as outside the boundaries of everyday politics. Yet, in Weimar Germany before its collapse and in contemporary South Asian societies, they have been associated with the routine politics of public rallies, demonstrations, and elections. In Weimar Germany, they were associated with a movement that ultimately brought down the regime and replaced it with another. In Sri Lanka in 1983 they proved to be the prelude to a prolonged civil war. However, in India, riots and other forms of localized violence appear to have been integrated into routine politics. They are anticipated by-products of election campaigns and mass mobilizations that precede them. They are associated with movements by caste and communal groups to assert their rights or maintain their privileges against challengers.

Perhaps, then, we should consider riots themselves as part of the traditional repertoire of political action, of both routine politics and movement politics. Or should we consider riots as older forms of collective action that have been infused with new meanings?¹³ While there is ample evidence that contemporary riots in India and elsewhere have different meanings from those ascribed to them by Hobsbawm, Rude, and Davis, many traditional practices seem common between riots in sixteenth-century France and twentieth-century India. Riots, it would seem, are among the most versatile forms of collective action, capable of carrying many meanings in different times and places, occurring as part of routine as well as movement politics, and displaying a multiplicity of ritualized practices.

For example, much of what Tarrow describes as the old repertoire of collective action appears in riots in India today. Just as Catholics burned Protestant churches with the parishioners inside them,¹⁴ so Hindu rioters attack Muslims seeking refuge in their mosques, burn the houses of Muslims with their occupants inside, and vandalize or destroy their mosques. Similarly, funerals and funeral processions today in India, as in nineteenth-century Europe,¹⁵ provide occasions for riots and massacres. More than two thousand Sikhs were massacred in Delhi in 1984 in the anti-Sikh riots that followed Mrs. Gandhi’s assassination, beginning during her cremation and continuing for some days afterwards.

However, unlike other forms of collective action in which the action itself is designed to convey precisely the demand, riots involve what Stanley J. Tambiah calls focalization and transvaluation of meaning.¹⁶ Riots are supposed to appear spontaneous, rather than planned, even when, as in the case of Nazi Germany, they are precisely organized, directed, and limited to specific targets. Moreover, there is in fact much that occurs during most riots that is not preplanned and precisely organized. Riots, therefore, are either designed to convey or their meaning must be captured to convey a diffuse message about alleged popular feelings and how they may be used against the authorities or one’s opponents or scapegoats. They are a warning to one’s ethnic and political rivals. It is here that the struggle for representation and meaning comes in, which most distinguishes riots from other forms of collective action. Riots, therefore, are partly organized, partly spontaneous forms of collective action designed to appear or made to appear afterwards as spontaneous expressions of popular feeling. In the hands of social scientists, they may even be transformed after the fact into rebellions, insurgencies, insurrections, and revolutions.

Riots, it is argued here, like other forms of collective action, have characteristic forms of organization and leadership,¹⁷ sometimes occur in waves that appear similar to sustained campaigns or movement cycles, and also display a characteristic repertoire of collective violence.¹⁸ Riots, too, like other forms of collective action, have become modernized. They must have a broader meaning that goes beyond the particular acts of violence associated with them. They must be interpreted and represented. Their meaning must be made clear, though in divided societies—and societies in which riots are endemic are virtually by definition divided—their meaning will certainly be contested.

In some societies, such as India, riots are transformed into routine and ordinary events even while they are deplored and condemned as extraordinary threats to public order. Moreover, just as the press in revolutionary America made rebellion ordinary,¹⁹ so does the press in India make riots ordinary. Just as Philadelphians could read about rebellion in the North in New York papers, so riot news is conveyed in India through both the sedate English-language presses and the less sedate and sometimes scurrilous vernacular presses. But India now has the more modern media as well, including both radio and the spreading availability of TV, so that the news of riots and news precipitating riots may be spread rapidly by the mightiest and most prestigious of all news transmitters, the BBC, and lately by CNN as well.

In some cities and towns, riots are endemic. I have argued above that sociological and ecological analyses do not provide an adequate basis for distinguishing those that are riot-prone from those that are not. What does invariably distinguish such cities and towns from others is the presence of what I have called above an institutionalized riot system. However, the forms of organization and leadership that characterize riots are different from those described by Tarrow for other forms of collective action.²⁰ They are not open and formalized. Regular meetings are not held. Leaderships are not elected. Memberships are fluid. There are, nevertheless, known persons who perform leadership roles, pools of persons from whom riot actors are drawn, and established links of communication.

Some of the leadership and communication roles are evident, for example, in descriptions of the Delhi Riots of 1984. Local members of the ruling party, including among them persons who had occupied and have since occupied high positions in the party organization and even in ministerial positions in the central government, provided direction, "mobilized their local clients and thugs, provided them with liquor, and directed sellers of kerosene oil, whose sale was restricted to permit holders, to distribute the fuel for arson. They also provided information about the targets—Sikh houses, business establishments, schools, and gurudwaras"²¹ Such information in turn was provided by informers and collaborators within the residential colonies selected for attack. The riot squads themselves were recruited from nearby castes, communities, and villages, from groups whose members have in the past done, or who are known to be available for, such work. Transportation for the riot squads also had to be provided, which meant recruiting bus drivers and vehicles from the Delhi Transportation Corporation, a public corporation. To start the riot and spread it, to provide noise, cover, and additional spontaneous recruits, further riot specialists also came into play: specially designated forces from the riot squads to act as mobile gangs, persons designated to spread rumors, shouters of slogans and instigators of violence among the public.²² There are, then, a whole series of specialized roles that are occupied in larger riots, including provocateurs, monitors, informers, riot captains and thugs,²³ provisioners of transport and liquor, criminals, bomb manufacturers, journalists and pamphleteers, graffiti writers, and distributors and plasterers of scurrilous posters.

It is evident from most accounts that formal organizations such as the ruling party and its party offices and public employees and public organizations were involved in the Delhi Riots of 1984, but not as such. That is to say, there exist in effect in Delhi lists of persons who form a loosely organized network that can be called up quickly when the need arises, when it is desired to produce a riot. Such networks exist elsewhere in India and, I believe, have existed in Russia in the nineteenth century and in many other places, including cities in the United States today.

Although it would be a mistake to imagine that all riots are like those in Delhi in 1984 and that they all involve such deliberate planning, I believe that all riot-prone cities and towns do have to a greater or lesser degree such informal organizational networks. They also have something else, which is central to the notion of institutionalized riot systems, namely, a network of persons who maintain communal, racial, and other ethnic relations in a state of tension, of readiness for riots. Here I part company definitively with the sequence theorists of collective action, who imagine a state of tension arising out of grievances, frustrations, and discriminations in the relations between two communities, which requires only the proverbial spark to ignite it. On the contrary, there are regular fire-tenders who maintain the fuel at a combustible level, sometimes stoking it, sometimes letting it smolder. They are the conversion specialists, who know how to convert a moment of tension into a grander, riotous event.

These fire-tenders occupy formal and informal roles in existing organizations and outside them. In the most riot-prone cities, one is certain to find community organizations, cultural organizations, and/or political parties devoted to the advancement of their community, who also depict members of another, designated community as either an enemy, an oppressor, or a threat to their own community. Such organizations in India include especially the RSS and the BJP,²⁴ the former a Hindu cultural organization and the latter a party of militant Hindu nationalism, whose leading positions are occupied by RSS-persons. Other organizations in the RSS family²⁵ include the Vishwa Hindu Parishad (VHP)²⁶ and the Bajrang Dal.²⁷ The former organization is devoted to mobilizing the Hindu community in India to rectify alleged past wrongs committed against Hindus by Muslims, notably the construction of mosques upon temples allegedly destroyed for the purpose. The Bajrang Dal is a fighting protection squad for the other organizations, a somewhat pathetic, but nevertheless dangerous, version of the Nazi SA.

In other towns in some parts of India there are Muslim organizations of this type, but they are far less widespread and less well organized, disciplined, and directed. There are other organizations as well, most of whose members are honestly devoted primarily to cultural activities and/or religious reform within their communities, whose members also may nevertheless play roles, deliberate or inadvertent, in riot systems. Such organizations include the Arya Samaj among Hindus in north India. Some towns known to be riot-prone do not necessarily have a full panoply of organized groups or the groups may be in an early stage of organization. But they all have individuals, formally affiliated or informally associated with such organizations or the ideas for which they stand, who play the roles of fire-tenders.

The role of fire-tender, moreover, is not always one of ignition. Most often, in fact, it is one of merely tending the flame and preventing its ignition when the circumstances are not right for a riot. Indeed, some of these fire-tenders masquerade as promoters of communal peace and harmony. They may even be members of local peace committees. In whatever role they are playing at the moment, there is nevertheless a constant purpose, namely, to maintain an awareness at all levels of society of the potential for a riot. In India, such persons are, for example, immediately informed whenever a Hindu girl elopes with a Muslim boy, a cow dies under circumstances that appear suspicious and when Muslims are involved, a Hindu idol miraculously appears in some public place or on some site occupied by Muslims, a Hindu is injured or rumored to have been injured or killed in a brawl with a Muslim, a well-off Muslim buys property in a predominantly Hindu area, and so on and on.

When, then, are the circumstances right for a riot? These circumstances differ in different countries at different times, but there is one that is common in party-electoral polities, namely, before and during elections and during movements of mass mobilization, especially when the political balance between contending forces appears to be changing, that is, when political opportunities are such that a riot may increase public sympathy for one’s own party or movement and weaken one’s rivals. The same kinds of circumstances also may determine the selection of otherwise everyday, trivial, and isolated incidents or small riots for magnification and transvaluation.

The roles of fire-tenders are crucial to the maintenance of riot systems, but to revert to analysis of the system as a whole and change the metaphor, the view presented here is that riots are best seen as dramatic productions with large casts of extras. They are productions, in other words, which are partly organized and in which the extras are cleverly used or, so to say, in which extensive ad-libbing occurs in order to convey the impression of spontaneity.²⁸ When a riot goes off,²⁹ then very large numbers of extras may indeed become involved. Even some of these, however, may be eagerly waiting for the signal to play their part, to come on stage. That signal may come in the form of provocative shouts and slogans to incite a crowd to attack or from the hurling of brickbats or the sound of breaking glass. Everyone knows his part or makes his own part as the scenes develop.

It is widely believed in India, and it is also one of the most common beliefs about riots in most places, that they are nearly always urban phenomena. Indeed, in India this demographic observation is commonly followed by a sigh of relief, suggesting many assuaging beliefs and feelings about them, to wit: The villagers, at least, have not been corrupted by the tensions and turmoils of urban life; bad relations between our peoples of different religions are not endemic, not built into the prejudices of the people, but are manufactured in the cities; the country is safe since these riots are contained in what is, after all, only a small part of it. Sometimes this demographic observation is accompanied by popular sociological explanations, which repeat much of the academic sociology of riots: They are consequences of urbanization; of uprooting and migration of rural people into the cities, juxtaposing peoples of different castes and communities in unfamiliar surroundings in a dog-eat-dog, competitive economic environment; of political mobilizations made possible by the existence of large masses of common people.

In fact, however, here as elsewhere in both popular and academic accounts of riots, this belief, if not simply false, is a function of definition, labeling, and the construction of meaning. If a riot is defined as requiring a minimum number of persons, as in most official riot acts, then most village brawls fall short of official requirements. But so do most urban brawls. In the cities, however, even a small brawl between a black and a white in America or a Hindu and a Muslim in India easily turns into a potential or an actual riot.

The differences between the two situations relate to the diffusion of information and to the relative importance of local knowledge. News of a barroom or teashop or street-corner brawl in the United States or India is transmitted rapidly in the cities by rumor, association of people in common organizations, print, and electronic media. The persons involved in the incident are known through such forms of transmission only as categories, not as real persons. In an Indian village, in contrast, the persons are known, the cause of their conflict is often also obvious, their group affiliations and the numbers of supporters each is likely to be able to produce in a broader conflict are also known, as are the consequences for village life if the brawl is allowed to expand. These are factors that make for containment rather than expansion, but they do not prevent the frequent outbreak of violent brawls and killings.

Large-scale riotous events are also easier